G Model
JINJ-5655; No. of Pages 5 Injury, Int. J. Care Injured xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Injury journal homepage: www.elsevier.com/locate/injury
The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery Sameer K. Khan a,*, Stephen P. Rushton b, David W. Shields a, Kenneth G. Corsar a, Ramsay Refaie a, Andrew C. Gray a, David J. Deehan a a b
Royal Victoria Infirmary, Newcastle upon Tyne University Hospitals NHS Trust, Newcastle upon Tyne, United Kingdom School of Biology, Newcastle University, Newcastle upon Tyne, United Kingdom
A R T I C L E I N F O
A B S T R A C T
Article history: Accepted 16 February 2014
Introduction: 30-day mortality is routinely used to assess proximal femoral fracture care, though patients might remain at risk for poor outcome for longer. This work has examined the survivorship out to one year of a consecutive series of patients admitted for proximal femoral fracture to a single institution. We wished to quantify the temporal impact of fracture upon mortality, and also the influence of patient age, gender, surgical delay and length of stay on mortality from both cardiorespiratory and noncardiorespiratory causes. Patients and methods: Data were analysed for 561 consecutive patients with 565 fragility type proximal femoral fractures treated surgically at our trauma unit. Dates and causes of death were obtained from death certificates and also linked to data from the Office of National Statistics. Mortality rates and causes were collated for two time periods: day 0–30, and day 31–365. Results: Cumulative incidence analysis showed that mortality due to cardiorespiratory causes (pneumonia, myocardial infarction, cardiac failure) rose steeply to around 100 days after surgery and then flattened reaching approximately 12% by 1 year. Mortality from non-cardiorespiratory causes (kidney failure, stroke, sepsis etc.) was more progressive, but with a rate half of that of cardiorespiratory causes. Progressive modelling of mortality risks revealed that cardiorespiratory deaths were associated with advancing age and male gender (p < 0.001 for both), but the effect of age declined after 100 days. Non-cardiorespiratory deaths were not time-dependent. Conclusion: We believe this analysis extends our understanding of the temporal impact of proximal femoral fracture and its surgical management upon outcome beyond the previously accepted standard (30 days) and supports the use of a new, more relevant timescale for this high risk group of patients. It also highlights the need for planning and continuing physiotherapy, respiratory exercises and other chest-protective measures from 31 to 100 days. ß 2014 Elsevier Ltd. All rights reserved.
Keywords: Proximal femoral fractures Cardiorespiratory mortality Medical certificate of cause of death Cumulative incidence analysis Pneumonia
Introduction Despite widespread implementation of national guidelines [1,2], improvements in care pathways and advances in surgical techniques, mortality after proximal femoral fractures remains high with rates of between 20% and 30% frequently quoted out to the first post operative year [3]. It is acknowledged that the majority of such deaths occur in the early post-operative period [4,5], with the National Hip Fracture Database reporting a 30-day mortality rate of just over 8% [6,7]. The arbitrary timeline of
* Corresponding author. Tel.: +44 7447030046; fax: +44 1916600801. E-mail address: [email protected] (S.K. Khan).
30 days after surgery is commonly cited in many clinical and health related outcome publications as a surrogate quality indicator for proximal femoral care provision [8–10]. However, there is little corroborative evidence to validate this particular time period or to reflect upon the temporal impact of this injury. Key non-modifiable factors are known to increase the risk of excessive mortality following proximal femoral fractures. These include age, male sex and anatomical position of the fracture, with a variable reported time period for increased mortality risk [11– 14]. Modifiable risk factors, such as chest infection and anaemia, might intuitively be expected to have a greater impact perioperatively because of the stress imposed by the fracture and its surgical treatment. However with time, such influences will lessen and the outcome should ultimately not be influenced by the inpatient
http://dx.doi.org/10.1016/j.injury.2014.02.024 0020–1383/ß 2014 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Khan SK, et al. The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery. Injury (2014), http://dx.doi.org/10.1016/j.injury.2014.02.024
G Model
JINJ-5655; No. of Pages 5 2
S.K. Khan et al. / Injury, Int. J. Care Injured xxx (2014) xxx–xxx
episode. Accurate delineation of the duration of impact of trauma from injury and the following surgical procedure will allow the clinician to: (1) assess more accurately and reliably the impact of early intervention measures upon outcome and (2) separate the effects of intervention from the influence of high level of comorbidities associated with the natural history of these fractures. This work has examined the survivorship out to one year of a consecutive series of patients admitted for proximal femoral fracture to a single institution. We aimed to quantify the temporal impact of the fracture upon mortality after surgery and in the context of mortality due to unrelated causes. We hypothesised that mortality after proximal femoral surgery was dependent on patient age, gender, surgical delay and length of stay. We tested a further hypothesis that mortality from cardiorespiratory causes would be higher perioperatively and that it would be dependent on patient age, time to surgery and length of stay. Patients and methods After approval from the hospital’s research ethics board, a retrospective chart review was conducted on consecutive patient episodes with fragility-type proximal femoral fractures admitted from the emergency department of a tertiary referral teaching hospital over a 24-month period (December 2008–December 2010) and treated surgically at its trauma unit. All patients with non-fragility type (pathological) fractures (n = 10) were transferred to an off-site elective unit for definitive treatment and were excluded from this study, as were patients treated non-operatively (n = 19). The day-to-day medical management of all patients was supervised by orthogeriatricians. Cardiology and respiratory consultations were sought for patients when requested by the orthogeriatricians or anaesthetists. Uniform blood transfusion and antibiotic policies were in place at the hospital during the study period. Data on patient demographics, dates and times of presentation to the emergency department, and admission to and discharge from the trauma ward and the hospital were obtained from the hospital’s patient management system. Times to surgery (non-normally distributed) and lengths of hospital stay (normally distributed) were calculated for all patient episodes [15]. Dates of in-hospital deaths were obtained from the inpatient management system. Immediate and/or underlying causes of death were identified from the registers containing the medical certificates of cause of death (MCCD) in the hospital. Data on dates and causes of out-of-hospital deaths were obtained from the records of their respective registered family physicians’ practices. For any remaining unaccounted deaths, the patient identifiers were linked to the ‘mortality file’ generated by the Office of National Statistics [ONS; Cardiff, UK]. This file contains data extracted from actual MCCDs filled in by general practitioners. All mortalities and respective causation were collated for all patients to a censoring point of one year after surgery. The recorded causes of death were referenced against the ICD-10 compendium to ensure each type contains system-specific diagnoses only. To facilitate analysis, the documented causes of death were grouped as follows: 1. Respiratory causes: ‘Pneumonia’, ‘Respiratory failure’, ‘COPD’. 2. Cardiac causes: ‘Myocardial infarction’, ‘Cardiac failure’, ‘Ischaemic heart disease’. 3. Renal causes: ‘Acute kidney failure’, ‘Chronic kidney disease’. 4. Stroke: ‘Stroke’, ‘Cerebrovascular disease’. 5. Acute abdominal events: ‘Visceral perforation’, ‘Intestinal haemorrhage’, ‘Intestinal ischaemia’. 6. Sepsis: ‘Multiorgan failure from sepsis/urinary infection’.
7. Dementia. 8. Frailty of old age. 9. Malignancy: ‘Neoplasm’.
Statistical methods Mortality rates and causes were collated for two time periods: period 1 (from surgery to 30 days) and period 2 (from 31 to 365 days after surgery). In the first instance these were tabulated and numbers of deaths due to different causes were compared simplistically using Fisher’s exact test. For some causes there were few cases so causes of death were regrouped into two major categories: group 1 (cardiorespiratory) and group 2 (noncardiorespiratory). We used cumulative incidence analysis to illustrate mortality due to the two groups of causes. A progressive modelling strategy was then produced to investigate risks for mortality from the different causes. The cause-specific mortality of the two groups was examined so as to identify risk factors for mortality assuming that risks for mortality were independent for the two groups. We then relaxed the assumption of independence of risk and used competing risks regression to investigate risk factors for death after surgery assuming that the risk factors impacted on more than one cause, using competing risks regression. We assessed the extent to which the effects of the individual variables impacted on combined cardiac and respiratory mortality were proportional i.e. did not change. This was estimated through a regression analysis of the residuals of the model against time, in a manner analogous to the residual analysis of Therneau and Grambsch used to analyse time dependency in Cox proportional hazards models [16]. Results Five hundred and sixty one patients (137 males, 424 females) were admitted and treated surgically for 565 fractures (254 extracapsular, 311 intracapsular). Five patients suffered fractures on both sides but separated in time, i.e. there were no admissions with simultaneous bilateral fractures. The unit of analysis is therefore the patient episode, instead of the patient. The mean age was 80.6 years (median 82; range 65–100 years). The median and mode ASA grade was 3 (with 63% of the sample at the modal value). The median surgical delay for the cohort was 22 h 35 min (range: 3 h 10 min–434 h 26 min). The mean (SD) post-operative length of stay on the trauma ward was 17 days (16.3) while the total postoperative stay in hospital averaged 30 days (30.3). The cumulative 30-day mortality was 5.3% (n = 30), which increased to 19.2% (n = 78) at one year. When grouped by fracture type (extracapsular v/s intracapsular), the mortality rates were similar at 30 days (5.46% v/s 5.49%, p = 1) but more of the intracapsular fracture patients had died in period 2 (10.8% v/s 21.1%, p = 0.03). One hundred and three episodes (18.2%) were associated with a delay to surgery of more than 36 h, 50 of these were for various medical reasons. Mortality rates for periods 1 and 2 were not significantly different between episodes with and without delays to surgery (p = 0.3 and p = 1, respectively). Fig. 1 illustrates the causes of death for the two time periods. Cardiorespiratory causes accounted for the majority of deaths in both periods (70% and 57.6%, respectively). Pneumonia contributed directly to 40% of deaths in both periods. There were more deaths due to dementia, malignancy and frailty over period 2 than in period 1. Cumulative incidence functions for the cardiorespiratory mortality compared to non-cardiorespiratory causes are shown in Fig. 2. Cumulative incidence of mortality due to cardiorespiratory causes rose steeply to around 100 days after surgery and then flattened reaching approximately 12% by one
Please cite this article in press as: Khan SK, et al. The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery. Injury (2014), http://dx.doi.org/10.1016/j.injury.2014.02.024
G Model
JINJ-5655; No. of Pages 5 S.K. Khan et al. / Injury, Int. J. Care Injured xxx (2014) xxx–xxx
3
Fig. 1. Causes of death recorded for patients dying in the two time periods.
year. The non-cardiorespiratory mortality was seen to be more progressive, but with a rate of approximately 50% of that demonstrated in the cardiorespiratory category. Cause-specific mortality for cardiorespiratory causes (assuming independence relative to other cause mortality) was associated with age at admission (z = 3.41, p < 0.001) and gender, with females at lowest risk (z = 3.63, p < 0.001). Neither surgical delay nor length of stay was a significant risk factor for mortality. This indicated that these patient management issues were not significantly implicated in subsequent mortality. For non-cardiorespiratory mortality, only age was a significant risk factor (z = 3.92, p < 0.001). The risk of cardiorespiratory deaths associated with increased age, but the hazard of death from this cause was not constant in the period following surgery as a chi squared test of the assumption of constant hazard (termed proportionality) was significant (chi = 4.04, p = 0.04). The extent of non-proportionality can be demonstrated by considering the sum of residuals of the
fitted model and the estimated regression coefficients at each point in time. These are an approximation to the value of the regression coefficient at each event and a measure of how the hazard of dying with increased age changes with time. This is effectively a measure of this time dependency in the relationship between hazard of mortality and the respective covariate [16]. A plot of these time dependent coefficients (Fig. 3) with associated confidence limits (dotted lines) indicates that the effect of age at admission as a risk factor for mortality after fracture remained positive and significant to around 100–120 days after surgery, where the confidence intervals overlap with zero (no effect) on the coefficient (y) axis. This means that any contribution of age to increased risk of death from cardiorespiratory causes has declined by this point after fracture. There was no such time dependence in the analysis of non-cardiorespiratory causes. This would indicate that the risk of dying was associated with age at admission and stayed the same through the study period for this cohort of patients.
Fig. 2. Cumulative incidence of death out to one year from cardiac and respiratory failure and other causes following surgery for neck of femur fractures. Days since surgery shown along horizontal axis.
Fig. 3. Plot of the time dependent coefficients for the effect of age on risk of mortality from cardiac and respiratory failure. A lowess smoother (and CI) crosses zero point at after 120 days.
Please cite this article in press as: Khan SK, et al. The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery. Injury (2014), http://dx.doi.org/10.1016/j.injury.2014.02.024
G Model
JINJ-5655; No. of Pages 5 S.K. Khan et al. / Injury, Int. J. Care Injured xxx (2014) xxx–xxx
4
In the competing risk regression analysis, age at admission (z = 3.14, p = 0.002) and gender (z = 3.68, p < 0.001) were significant risk factors for mortality. There was also a significant negative relationship between the residuals of this model and time (t = 2.84, p = 0.006) suggesting that the effect of age at admission and gender on risk were not consistent through time and that the model underestimated mortality in period 1. Discussion There were four key findings in this study. Firstly, respiratory and cardiac causes are the predominant causes of death at both 30 days and at one year. Secondly, the risk of death from cardiorespiratory causes is age- and gender-dependent with elderly men being most at-risk, however, the influence of age declines decidedly after 100 days. Thirdly, surgical delay and length of stay do not significantly influence cardiorespiratory mortality. Finally, the role of age in noncardiorespiratory causes of death is neither time-related, nor gender-dependent. This work thus suggests that the first 100 days after surgery represent the ‘at-risk’ period for poor outcome from exclusively cardiorespiratory causes. Mortality analysis in proximal femoral fractures is inherently complex. The heterogeneity of the case mix, high level of comorbidity and variations in timing of surgery have been shown to exhibit a high degree of inter-dependence, thereby confounding simple interpretations. Previous work has relied upon risk stratification and multivariable analyses to define prognostic variables [17–19]. Whilst useful, these do not confidently explain the degree to which these variables influence each another and how they interact to predict outcome [9]. Previous studies also report standardised mortality rates in fracture patients compared to the general population. In practice however, it is difficult to obtain an equivalent age-, sex- and comorbidity-matched cohort that really does match the proximal femoral fracture population, and any formal arbitrary comparison could lead to both type 1 and 2 errors. This persuaded us to use a robust mathematical modelling technique to evaluate these complex interactions. The cohort is sufficiently large and is consistent with previous studies reporting causes of death in these patients [4,20]. We accept that there are certain limitations to this work, e.g. the absence of data on pre-operative comorbidities, the reliance on MCCDs for causes of death, and the seemingly simplistic grouping scheme adopted to classify the causes of death. We have however used ICD codes which have been validated to be sensitive and specific in identifying respiratory and coronary heart disease [21]. These are also used by the ONS to select the underlying cause for routine mortality statistics from MCCDs reported by nationwide general practices. At our unit, all hospital deaths occurring within 30 days of surgery are routinely discussed with the coroner, however only a few post-mortems were carried out at the discretion of the coroner. Post-mortems on all patients dying within one year of proximal femoral fractures are not possible. Acharya and co workers have previously reported a strong concordance (in terms of frequency and spectrum) between causes of death recorded by medical practitioners and coroners for hospital deaths following proximal femoral fractures [22]. We also do not have data on the types of anaesthesia used in different patients, though it might be argued that the direct immunomodulatory effect of the anaesthetic per se may have passed beyond 30 days. The three most frequent causes of death identified in this study were pneumonia (39%), myocardial infarction (9.2%) and cardiac failure (8.3%), with mortality rates comparable to those previously reported [17,23,24]. These cardiorespiratory complications have previously been reported to occur more commonly in the first 30 days, but the exact time point where their incidence reaches a
plateau has not been elucidated to date. Advancing age increases the incidence of these complications [11,25,26] as well as the risk of mortality from these [27]. This analysis adds to existing knowledge in two ways. It defines the duration of the at-risk period during which age continues to cause significantly high mortality from cardiorespiratory causes. Secondly, it identifies the patient subgroup most at-risk of dying from these causes in this period. In the presented modelling analysis, mortality did not seem to be influenced by surgical delay and length of stay. Repeated audit cycles performed within our own unit strongly suggest that medically fit patients rarely have a delay to surgery and that the vast majority are delayed with good reason due to the need for medical optimisation. Some authors have previously reported increased mortality with extracapsular fractures [12], whilst recent studies have shown comparable age-adjusted mortality for the two fracture types [28]. Potential confounders include age, comorbidities and pre- and post-operative mobility and residential status etc. [29]. This study has found an excess mortality with intracapsular fractures, suggesting that the relationship between fracture type and excess mortality is unclear. The impact of age on mortality abates after the 100-day vulnerable period. This might reflect the declining impact of the surgery on perioperative survival, through normalizing physiology and reducing incidence of chest infections and cardiac events. It thus represents a more accurate measure of patient survival following proximal femoral fracture than the 30-day mortality index used in orthopaedic trauma studies. This is similar to the 90day period currently being proposed as a post-operative performance indicator in cancer patients, a population with comparable prognosis [30]. The minimum inpatient stay for the majority of proximal femoral fracture patients is at least three weeks [31]. This is adequate time to plan continued physiotherapy, especially chest physiotherapy, for the 31–100 day period. This study extends our understanding of the temporal impact of surgical management of proximal femoral fracture upon outcome and supports the use of a new more relevant timescale for this high-risk group of patients. Conflict of interest statement The authors declare that they have no conflict of interest in connection with this paper. No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. References [1] Scottish Intercollegiate Guidelines Network. Management of hip fracture in older people, a national clinical guideline. (SIGN publication no. 111) 2009. http://www.sign.ac.uk/guidelines/fulltext/111/index.html.(date last accessed 15 January 2013). [2] National Institute for Health and Clinical Excellence. National Institute for Health and Clinical Excellence. The management of hip fracture in adults (Clinical guideline CG124); 2011 , http://www.guidance.nice.org.uk/CG124 (date last accessed 15 August 2012). [3] Haleem S, Lutchman L, Mayahi R, Grice JE, Parker MJ. Mortality following hip fracture: trends and geographical variations over the last 40 years. Injury 2008;39:1157–63. [4] Pioli G, Barone A, Giusti A, Oliveri M, Pizzonia M, Razzano M, Palummeri E. Predictors of mortality after hip fracture: results from 1-year follow-up. Aging Clin Exp Res 2006;18:381–7. [5] Tosteson AN, Gottlieb DJ, Radley DC, Fisher ES, Melton 3rd LJ. Excess mortality following hip fracture: the role of underlying health status. Osteoporos Int 2007;18:1463–72. [6] National Hip Fracture Database. The national hip fracture database national report 2012, http://www.nhfd.co.uk/003/hipfractureR.nsf/luMenuDefinitions/CA920122A244F2ED802579C900553993/$file/NHFD%20National%20Report%202012.pdf?OpenElement (date last accessed 15 January 2013). [7] National Hip Fracture Database. The national hip fracture database preliminary national report 2009, http://www.nhfd.co.uk/003/hipfracturer.nsf/6ec433ed9efaa78e802572e3003a3517/ed40fa45ba9877108025779f0041fbca/$FILE/Preliminary%20National%20Report.pdf (date last accessed 15 January 2013).
Please cite this article in press as: Khan SK, et al. The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery. Injury (2014), http://dx.doi.org/10.1016/j.injury.2014.02.024
G Model
JINJ-5655; No. of Pages 5 S.K. Khan et al. / Injury, Int. J. Care Injured xxx (2014) xxx–xxx [8] Nielsen KA, Jensen NC, Jensen CM, Thomsen M, Pedersen L, Johnsen SP, Ingeman A, Bartels PD, Thomsen RW. Quality of care and 30 day mortality among patients with hip fractures: a nationwide cohort study. BMC Health Serv Res 2009;9:186. [9] Wiles MD, Moran CG, Sahota O, Moppett IK. Nottingham hip fracture score as a predictor of one year mortality in patients undergoing surgical repair of fractured neck of femur. Br J Anaesth 2011;106:501–4. [10] Kirkland LL, Kashiwagi DT, Burton MC, Cha S, Varkey P. The Charlson Comorbidity Index Score as a predictor of 30-day mortality after hip fracture surgery. Am J Med Qual 2011;26:461–7. [11] Jameson SS, Khan SK, Baker P, James P, Gray A, Reed MR, Deehan DJ. A national analysis of complications following hemiarthroplasty for hip fracture in older patients. QJM 2012;105:455–60. [12] Johnston AT, Barnsdale L, Smith R, Duncan K, Hutchison JD. Change in longterm mortality associated with fractures of the hip: evidence from the Scottish hip fracture audit. J Bone Joint Surg [Br] 2010;92:989–93. [13] Richmond J, Aharonoff GB, Zuckerman JD, Koval KJ. Mortality risk after hip fracture. J Orthop Trauma 2003;17:53–6. [14] Tsuboi M, Hasegawa Y, Suzuki S, Wingstrand H, Thorngren KG, Mortality. Mobility after hip fracture in Japan: a ten-year follow-up. J Bone Joint Surg [Br] 2007;89-B:461–6. [15] Khan SK, Jameson SS, Avery PJ, Gray AC, Deehan DJ. Does the timing of presentation of neck of femur fractures affect the outcome of surgical intervention. Eur J Emerg Med 2013;20:178–81. [16] Therneau TM, Grambsch PM. Modelling Survival Data: extending the Cox model. New York, NY: Springer-Verlag New York Inc.; 2000. [17] Roche JJ, Wenn RT, Sahota O, Moran CG. Effect of comorbidities and postoperative complications on mortality after hip fracture in elderly people: prospective observational cohort study. BMJ 2005;331:1374. [18] Hu F, Jiang C, Shen J, Tang P, Wang Y. Preoperative predictors for mortality following hip fracture surgery: a systematic review and meta-analysis. Injury 2012;43:676–85. [19] de Luise C, Brimacombe M, Pedersen L, Sørensen HT. Comorbidity and mortality following hip fracture: a population-based cohort study. Aging Clin Exp Res 2008;20:412–8.
5
[20] Perez JV, Warwick DJ, Case CP, Bannister GC. Death after proximal femoral fracture—an autopsy study. Injury 1995;26:237–40. [21] Harriss LR, Ajani AE, Hunt D, et al. Accuracy of national mortality codes in identifying adjudicated cardiovascular deaths. Aust N Z J Public Health 2011;35:466–76. [22] Acharya MR, Williams SC, Harper WM. Hip fracture patients: why do they die? J Bone Joint Surg [Br] 2006;88-B(Suppl I):39. [23] Wood DJ, Ions GK, Quinby JM, Gale DW, Stevens J. Factors which influence mortality after subcapital hip fracture. J Bone Joint Surg [Br] 1992;74: 199–202. [24] Yousri T, Tsai A, Thyagarajen D, Livingstone J. Medical causes of death in hip fragility fractures: mortality analysis in a primary care trauma centre. In: EFORT congress; 2011 [abstract]. [25] Cruz E, Cano JR, Benitez-Parejo N, Rivas-Ruiz F, Perea-Milla E, Guerado E. Age as a risk factor of nosocomial infection after hip fracture surgery. Hip Int 2010;20(Suppl 7):19–25. [26] National Confidential Enquiry into Patient Outcome and Death. Knowing the risk—a review of the peri-operative care of surgical patients, http://www.ncepod.org.uk/2011report2/downloads/POC_summary.pdf (date last accessed 15 January 2013). [27] Gupta BP, Kirkland LL, Huddleston JM, Huddleston PM, Burton MC, Dirk LR, et al. Risk factors for long-term mortality in very elderly after post-operative myocardial infarction following hip fracture surgery. J Am Coll Cardiol 2011;57:E1038. [28] Finnes TE, Meyer HE, Falch JA, Medhus AW, Wentzel-Larsen T, Lofthus CM. Secular reduction of excess mortality in hip fracture patients >85 years. BMC Geriatrics 2013;13:25. [29] Cornwall R, Gilbert MS, Koval KJ, Strauss E, Siu AL. Functional outcomes and mortality vary among different types of hip Fractures: a function of patient characteristics. Clin Orthop Relat Res 2004;425:64–71. [30] Damhuis RAM, Wijnhoven BPL, Plaisier PW, Kirkels WJ, Kranse R, van Lanschot JJ. Comparison of 30-day, 90-day and in-hospital postoperative mortality for eight different cancer types. Br J Surg 2012;99:1149–54. [31] Hahnel J, Burdekin H, Anand S. Re-admissions following hip fracture surgery. Ann R Coll Surg Engl 2009;91:591–5.
Please cite this article in press as: Khan SK, et al. The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery. Injury (2014), http://dx.doi.org/10.1016/j.injury.2014.02.024
JINJ-5655; No. of Pages 5 Injury, Int. J. Care Injured xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Injury journal homepage: www.elsevier.com/locate/injury
The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery Sameer K. Khan a,*, Stephen P. Rushton b, David W. Shields a, Kenneth G. Corsar a, Ramsay Refaie a, Andrew C. Gray a, David J. Deehan a a b
Royal Victoria Infirmary, Newcastle upon Tyne University Hospitals NHS Trust, Newcastle upon Tyne, United Kingdom School of Biology, Newcastle University, Newcastle upon Tyne, United Kingdom
A R T I C L E I N F O
A B S T R A C T
Article history: Accepted 16 February 2014
Introduction: 30-day mortality is routinely used to assess proximal femoral fracture care, though patients might remain at risk for poor outcome for longer. This work has examined the survivorship out to one year of a consecutive series of patients admitted for proximal femoral fracture to a single institution. We wished to quantify the temporal impact of fracture upon mortality, and also the influence of patient age, gender, surgical delay and length of stay on mortality from both cardiorespiratory and noncardiorespiratory causes. Patients and methods: Data were analysed for 561 consecutive patients with 565 fragility type proximal femoral fractures treated surgically at our trauma unit. Dates and causes of death were obtained from death certificates and also linked to data from the Office of National Statistics. Mortality rates and causes were collated for two time periods: day 0–30, and day 31–365. Results: Cumulative incidence analysis showed that mortality due to cardiorespiratory causes (pneumonia, myocardial infarction, cardiac failure) rose steeply to around 100 days after surgery and then flattened reaching approximately 12% by 1 year. Mortality from non-cardiorespiratory causes (kidney failure, stroke, sepsis etc.) was more progressive, but with a rate half of that of cardiorespiratory causes. Progressive modelling of mortality risks revealed that cardiorespiratory deaths were associated with advancing age and male gender (p < 0.001 for both), but the effect of age declined after 100 days. Non-cardiorespiratory deaths were not time-dependent. Conclusion: We believe this analysis extends our understanding of the temporal impact of proximal femoral fracture and its surgical management upon outcome beyond the previously accepted standard (30 days) and supports the use of a new, more relevant timescale for this high risk group of patients. It also highlights the need for planning and continuing physiotherapy, respiratory exercises and other chest-protective measures from 31 to 100 days. ß 2014 Elsevier Ltd. All rights reserved.
Keywords: Proximal femoral fractures Cardiorespiratory mortality Medical certificate of cause of death Cumulative incidence analysis Pneumonia
Introduction Despite widespread implementation of national guidelines [1,2], improvements in care pathways and advances in surgical techniques, mortality after proximal femoral fractures remains high with rates of between 20% and 30% frequently quoted out to the first post operative year [3]. It is acknowledged that the majority of such deaths occur in the early post-operative period [4,5], with the National Hip Fracture Database reporting a 30-day mortality rate of just over 8% [6,7]. The arbitrary timeline of
* Corresponding author. Tel.: +44 7447030046; fax: +44 1916600801. E-mail address: [email protected] (S.K. Khan).
30 days after surgery is commonly cited in many clinical and health related outcome publications as a surrogate quality indicator for proximal femoral care provision [8–10]. However, there is little corroborative evidence to validate this particular time period or to reflect upon the temporal impact of this injury. Key non-modifiable factors are known to increase the risk of excessive mortality following proximal femoral fractures. These include age, male sex and anatomical position of the fracture, with a variable reported time period for increased mortality risk [11– 14]. Modifiable risk factors, such as chest infection and anaemia, might intuitively be expected to have a greater impact perioperatively because of the stress imposed by the fracture and its surgical treatment. However with time, such influences will lessen and the outcome should ultimately not be influenced by the inpatient
http://dx.doi.org/10.1016/j.injury.2014.02.024 0020–1383/ß 2014 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Khan SK, et al. The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery. Injury (2014), http://dx.doi.org/10.1016/j.injury.2014.02.024
G Model
JINJ-5655; No. of Pages 5 2
S.K. Khan et al. / Injury, Int. J. Care Injured xxx (2014) xxx–xxx
episode. Accurate delineation of the duration of impact of trauma from injury and the following surgical procedure will allow the clinician to: (1) assess more accurately and reliably the impact of early intervention measures upon outcome and (2) separate the effects of intervention from the influence of high level of comorbidities associated with the natural history of these fractures. This work has examined the survivorship out to one year of a consecutive series of patients admitted for proximal femoral fracture to a single institution. We aimed to quantify the temporal impact of the fracture upon mortality after surgery and in the context of mortality due to unrelated causes. We hypothesised that mortality after proximal femoral surgery was dependent on patient age, gender, surgical delay and length of stay. We tested a further hypothesis that mortality from cardiorespiratory causes would be higher perioperatively and that it would be dependent on patient age, time to surgery and length of stay. Patients and methods After approval from the hospital’s research ethics board, a retrospective chart review was conducted on consecutive patient episodes with fragility-type proximal femoral fractures admitted from the emergency department of a tertiary referral teaching hospital over a 24-month period (December 2008–December 2010) and treated surgically at its trauma unit. All patients with non-fragility type (pathological) fractures (n = 10) were transferred to an off-site elective unit for definitive treatment and were excluded from this study, as were patients treated non-operatively (n = 19). The day-to-day medical management of all patients was supervised by orthogeriatricians. Cardiology and respiratory consultations were sought for patients when requested by the orthogeriatricians or anaesthetists. Uniform blood transfusion and antibiotic policies were in place at the hospital during the study period. Data on patient demographics, dates and times of presentation to the emergency department, and admission to and discharge from the trauma ward and the hospital were obtained from the hospital’s patient management system. Times to surgery (non-normally distributed) and lengths of hospital stay (normally distributed) were calculated for all patient episodes [15]. Dates of in-hospital deaths were obtained from the inpatient management system. Immediate and/or underlying causes of death were identified from the registers containing the medical certificates of cause of death (MCCD) in the hospital. Data on dates and causes of out-of-hospital deaths were obtained from the records of their respective registered family physicians’ practices. For any remaining unaccounted deaths, the patient identifiers were linked to the ‘mortality file’ generated by the Office of National Statistics [ONS; Cardiff, UK]. This file contains data extracted from actual MCCDs filled in by general practitioners. All mortalities and respective causation were collated for all patients to a censoring point of one year after surgery. The recorded causes of death were referenced against the ICD-10 compendium to ensure each type contains system-specific diagnoses only. To facilitate analysis, the documented causes of death were grouped as follows: 1. Respiratory causes: ‘Pneumonia’, ‘Respiratory failure’, ‘COPD’. 2. Cardiac causes: ‘Myocardial infarction’, ‘Cardiac failure’, ‘Ischaemic heart disease’. 3. Renal causes: ‘Acute kidney failure’, ‘Chronic kidney disease’. 4. Stroke: ‘Stroke’, ‘Cerebrovascular disease’. 5. Acute abdominal events: ‘Visceral perforation’, ‘Intestinal haemorrhage’, ‘Intestinal ischaemia’. 6. Sepsis: ‘Multiorgan failure from sepsis/urinary infection’.
7. Dementia. 8. Frailty of old age. 9. Malignancy: ‘Neoplasm’.
Statistical methods Mortality rates and causes were collated for two time periods: period 1 (from surgery to 30 days) and period 2 (from 31 to 365 days after surgery). In the first instance these were tabulated and numbers of deaths due to different causes were compared simplistically using Fisher’s exact test. For some causes there were few cases so causes of death were regrouped into two major categories: group 1 (cardiorespiratory) and group 2 (noncardiorespiratory). We used cumulative incidence analysis to illustrate mortality due to the two groups of causes. A progressive modelling strategy was then produced to investigate risks for mortality from the different causes. The cause-specific mortality of the two groups was examined so as to identify risk factors for mortality assuming that risks for mortality were independent for the two groups. We then relaxed the assumption of independence of risk and used competing risks regression to investigate risk factors for death after surgery assuming that the risk factors impacted on more than one cause, using competing risks regression. We assessed the extent to which the effects of the individual variables impacted on combined cardiac and respiratory mortality were proportional i.e. did not change. This was estimated through a regression analysis of the residuals of the model against time, in a manner analogous to the residual analysis of Therneau and Grambsch used to analyse time dependency in Cox proportional hazards models [16]. Results Five hundred and sixty one patients (137 males, 424 females) were admitted and treated surgically for 565 fractures (254 extracapsular, 311 intracapsular). Five patients suffered fractures on both sides but separated in time, i.e. there were no admissions with simultaneous bilateral fractures. The unit of analysis is therefore the patient episode, instead of the patient. The mean age was 80.6 years (median 82; range 65–100 years). The median and mode ASA grade was 3 (with 63% of the sample at the modal value). The median surgical delay for the cohort was 22 h 35 min (range: 3 h 10 min–434 h 26 min). The mean (SD) post-operative length of stay on the trauma ward was 17 days (16.3) while the total postoperative stay in hospital averaged 30 days (30.3). The cumulative 30-day mortality was 5.3% (n = 30), which increased to 19.2% (n = 78) at one year. When grouped by fracture type (extracapsular v/s intracapsular), the mortality rates were similar at 30 days (5.46% v/s 5.49%, p = 1) but more of the intracapsular fracture patients had died in period 2 (10.8% v/s 21.1%, p = 0.03). One hundred and three episodes (18.2%) were associated with a delay to surgery of more than 36 h, 50 of these were for various medical reasons. Mortality rates for periods 1 and 2 were not significantly different between episodes with and without delays to surgery (p = 0.3 and p = 1, respectively). Fig. 1 illustrates the causes of death for the two time periods. Cardiorespiratory causes accounted for the majority of deaths in both periods (70% and 57.6%, respectively). Pneumonia contributed directly to 40% of deaths in both periods. There were more deaths due to dementia, malignancy and frailty over period 2 than in period 1. Cumulative incidence functions for the cardiorespiratory mortality compared to non-cardiorespiratory causes are shown in Fig. 2. Cumulative incidence of mortality due to cardiorespiratory causes rose steeply to around 100 days after surgery and then flattened reaching approximately 12% by one
Please cite this article in press as: Khan SK, et al. The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery. Injury (2014), http://dx.doi.org/10.1016/j.injury.2014.02.024
G Model
JINJ-5655; No. of Pages 5 S.K. Khan et al. / Injury, Int. J. Care Injured xxx (2014) xxx–xxx
3
Fig. 1. Causes of death recorded for patients dying in the two time periods.
year. The non-cardiorespiratory mortality was seen to be more progressive, but with a rate of approximately 50% of that demonstrated in the cardiorespiratory category. Cause-specific mortality for cardiorespiratory causes (assuming independence relative to other cause mortality) was associated with age at admission (z = 3.41, p < 0.001) and gender, with females at lowest risk (z = 3.63, p < 0.001). Neither surgical delay nor length of stay was a significant risk factor for mortality. This indicated that these patient management issues were not significantly implicated in subsequent mortality. For non-cardiorespiratory mortality, only age was a significant risk factor (z = 3.92, p < 0.001). The risk of cardiorespiratory deaths associated with increased age, but the hazard of death from this cause was not constant in the period following surgery as a chi squared test of the assumption of constant hazard (termed proportionality) was significant (chi = 4.04, p = 0.04). The extent of non-proportionality can be demonstrated by considering the sum of residuals of the
fitted model and the estimated regression coefficients at each point in time. These are an approximation to the value of the regression coefficient at each event and a measure of how the hazard of dying with increased age changes with time. This is effectively a measure of this time dependency in the relationship between hazard of mortality and the respective covariate [16]. A plot of these time dependent coefficients (Fig. 3) with associated confidence limits (dotted lines) indicates that the effect of age at admission as a risk factor for mortality after fracture remained positive and significant to around 100–120 days after surgery, where the confidence intervals overlap with zero (no effect) on the coefficient (y) axis. This means that any contribution of age to increased risk of death from cardiorespiratory causes has declined by this point after fracture. There was no such time dependence in the analysis of non-cardiorespiratory causes. This would indicate that the risk of dying was associated with age at admission and stayed the same through the study period for this cohort of patients.
Fig. 2. Cumulative incidence of death out to one year from cardiac and respiratory failure and other causes following surgery for neck of femur fractures. Days since surgery shown along horizontal axis.
Fig. 3. Plot of the time dependent coefficients for the effect of age on risk of mortality from cardiac and respiratory failure. A lowess smoother (and CI) crosses zero point at after 120 days.
Please cite this article in press as: Khan SK, et al. The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery. Injury (2014), http://dx.doi.org/10.1016/j.injury.2014.02.024
G Model
JINJ-5655; No. of Pages 5 S.K. Khan et al. / Injury, Int. J. Care Injured xxx (2014) xxx–xxx
4
In the competing risk regression analysis, age at admission (z = 3.14, p = 0.002) and gender (z = 3.68, p < 0.001) were significant risk factors for mortality. There was also a significant negative relationship between the residuals of this model and time (t = 2.84, p = 0.006) suggesting that the effect of age at admission and gender on risk were not consistent through time and that the model underestimated mortality in period 1. Discussion There were four key findings in this study. Firstly, respiratory and cardiac causes are the predominant causes of death at both 30 days and at one year. Secondly, the risk of death from cardiorespiratory causes is age- and gender-dependent with elderly men being most at-risk, however, the influence of age declines decidedly after 100 days. Thirdly, surgical delay and length of stay do not significantly influence cardiorespiratory mortality. Finally, the role of age in noncardiorespiratory causes of death is neither time-related, nor gender-dependent. This work thus suggests that the first 100 days after surgery represent the ‘at-risk’ period for poor outcome from exclusively cardiorespiratory causes. Mortality analysis in proximal femoral fractures is inherently complex. The heterogeneity of the case mix, high level of comorbidity and variations in timing of surgery have been shown to exhibit a high degree of inter-dependence, thereby confounding simple interpretations. Previous work has relied upon risk stratification and multivariable analyses to define prognostic variables [17–19]. Whilst useful, these do not confidently explain the degree to which these variables influence each another and how they interact to predict outcome [9]. Previous studies also report standardised mortality rates in fracture patients compared to the general population. In practice however, it is difficult to obtain an equivalent age-, sex- and comorbidity-matched cohort that really does match the proximal femoral fracture population, and any formal arbitrary comparison could lead to both type 1 and 2 errors. This persuaded us to use a robust mathematical modelling technique to evaluate these complex interactions. The cohort is sufficiently large and is consistent with previous studies reporting causes of death in these patients [4,20]. We accept that there are certain limitations to this work, e.g. the absence of data on pre-operative comorbidities, the reliance on MCCDs for causes of death, and the seemingly simplistic grouping scheme adopted to classify the causes of death. We have however used ICD codes which have been validated to be sensitive and specific in identifying respiratory and coronary heart disease [21]. These are also used by the ONS to select the underlying cause for routine mortality statistics from MCCDs reported by nationwide general practices. At our unit, all hospital deaths occurring within 30 days of surgery are routinely discussed with the coroner, however only a few post-mortems were carried out at the discretion of the coroner. Post-mortems on all patients dying within one year of proximal femoral fractures are not possible. Acharya and co workers have previously reported a strong concordance (in terms of frequency and spectrum) between causes of death recorded by medical practitioners and coroners for hospital deaths following proximal femoral fractures [22]. We also do not have data on the types of anaesthesia used in different patients, though it might be argued that the direct immunomodulatory effect of the anaesthetic per se may have passed beyond 30 days. The three most frequent causes of death identified in this study were pneumonia (39%), myocardial infarction (9.2%) and cardiac failure (8.3%), with mortality rates comparable to those previously reported [17,23,24]. These cardiorespiratory complications have previously been reported to occur more commonly in the first 30 days, but the exact time point where their incidence reaches a
plateau has not been elucidated to date. Advancing age increases the incidence of these complications [11,25,26] as well as the risk of mortality from these [27]. This analysis adds to existing knowledge in two ways. It defines the duration of the at-risk period during which age continues to cause significantly high mortality from cardiorespiratory causes. Secondly, it identifies the patient subgroup most at-risk of dying from these causes in this period. In the presented modelling analysis, mortality did not seem to be influenced by surgical delay and length of stay. Repeated audit cycles performed within our own unit strongly suggest that medically fit patients rarely have a delay to surgery and that the vast majority are delayed with good reason due to the need for medical optimisation. Some authors have previously reported increased mortality with extracapsular fractures [12], whilst recent studies have shown comparable age-adjusted mortality for the two fracture types [28]. Potential confounders include age, comorbidities and pre- and post-operative mobility and residential status etc. [29]. This study has found an excess mortality with intracapsular fractures, suggesting that the relationship between fracture type and excess mortality is unclear. The impact of age on mortality abates after the 100-day vulnerable period. This might reflect the declining impact of the surgery on perioperative survival, through normalizing physiology and reducing incidence of chest infections and cardiac events. It thus represents a more accurate measure of patient survival following proximal femoral fracture than the 30-day mortality index used in orthopaedic trauma studies. This is similar to the 90day period currently being proposed as a post-operative performance indicator in cancer patients, a population with comparable prognosis [30]. The minimum inpatient stay for the majority of proximal femoral fracture patients is at least three weeks [31]. This is adequate time to plan continued physiotherapy, especially chest physiotherapy, for the 31–100 day period. This study extends our understanding of the temporal impact of surgical management of proximal femoral fracture upon outcome and supports the use of a new more relevant timescale for this high-risk group of patients. Conflict of interest statement The authors declare that they have no conflict of interest in connection with this paper. No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. References [1] Scottish Intercollegiate Guidelines Network. Management of hip fracture in older people, a national clinical guideline. (SIGN publication no. 111) 2009. http://www.sign.ac.uk/guidelines/fulltext/111/index.html.(date last accessed 15 January 2013). [2] National Institute for Health and Clinical Excellence. National Institute for Health and Clinical Excellence. The management of hip fracture in adults (Clinical guideline CG124); 2011 , http://www.guidance.nice.org.uk/CG124 (date last accessed 15 August 2012). [3] Haleem S, Lutchman L, Mayahi R, Grice JE, Parker MJ. Mortality following hip fracture: trends and geographical variations over the last 40 years. Injury 2008;39:1157–63. [4] Pioli G, Barone A, Giusti A, Oliveri M, Pizzonia M, Razzano M, Palummeri E. Predictors of mortality after hip fracture: results from 1-year follow-up. Aging Clin Exp Res 2006;18:381–7. [5] Tosteson AN, Gottlieb DJ, Radley DC, Fisher ES, Melton 3rd LJ. Excess mortality following hip fracture: the role of underlying health status. Osteoporos Int 2007;18:1463–72. [6] National Hip Fracture Database. The national hip fracture database national report 2012, http://www.nhfd.co.uk/003/hipfractureR.nsf/luMenuDefinitions/CA920122A244F2ED802579C900553993/$file/NHFD%20National%20Report%202012.pdf?OpenElement (date last accessed 15 January 2013). [7] National Hip Fracture Database. The national hip fracture database preliminary national report 2009, http://www.nhfd.co.uk/003/hipfracturer.nsf/6ec433ed9efaa78e802572e3003a3517/ed40fa45ba9877108025779f0041fbca/$FILE/Preliminary%20National%20Report.pdf (date last accessed 15 January 2013).
Please cite this article in press as: Khan SK, et al. The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery. Injury (2014), http://dx.doi.org/10.1016/j.injury.2014.02.024
G Model
JINJ-5655; No. of Pages 5 S.K. Khan et al. / Injury, Int. J. Care Injured xxx (2014) xxx–xxx [8] Nielsen KA, Jensen NC, Jensen CM, Thomsen M, Pedersen L, Johnsen SP, Ingeman A, Bartels PD, Thomsen RW. Quality of care and 30 day mortality among patients with hip fractures: a nationwide cohort study. BMC Health Serv Res 2009;9:186. [9] Wiles MD, Moran CG, Sahota O, Moppett IK. Nottingham hip fracture score as a predictor of one year mortality in patients undergoing surgical repair of fractured neck of femur. Br J Anaesth 2011;106:501–4. [10] Kirkland LL, Kashiwagi DT, Burton MC, Cha S, Varkey P. The Charlson Comorbidity Index Score as a predictor of 30-day mortality after hip fracture surgery. Am J Med Qual 2011;26:461–7. [11] Jameson SS, Khan SK, Baker P, James P, Gray A, Reed MR, Deehan DJ. A national analysis of complications following hemiarthroplasty for hip fracture in older patients. QJM 2012;105:455–60. [12] Johnston AT, Barnsdale L, Smith R, Duncan K, Hutchison JD. Change in longterm mortality associated with fractures of the hip: evidence from the Scottish hip fracture audit. J Bone Joint Surg [Br] 2010;92:989–93. [13] Richmond J, Aharonoff GB, Zuckerman JD, Koval KJ. Mortality risk after hip fracture. J Orthop Trauma 2003;17:53–6. [14] Tsuboi M, Hasegawa Y, Suzuki S, Wingstrand H, Thorngren KG, Mortality. Mobility after hip fracture in Japan: a ten-year follow-up. J Bone Joint Surg [Br] 2007;89-B:461–6. [15] Khan SK, Jameson SS, Avery PJ, Gray AC, Deehan DJ. Does the timing of presentation of neck of femur fractures affect the outcome of surgical intervention. Eur J Emerg Med 2013;20:178–81. [16] Therneau TM, Grambsch PM. Modelling Survival Data: extending the Cox model. New York, NY: Springer-Verlag New York Inc.; 2000. [17] Roche JJ, Wenn RT, Sahota O, Moran CG. Effect of comorbidities and postoperative complications on mortality after hip fracture in elderly people: prospective observational cohort study. BMJ 2005;331:1374. [18] Hu F, Jiang C, Shen J, Tang P, Wang Y. Preoperative predictors for mortality following hip fracture surgery: a systematic review and meta-analysis. Injury 2012;43:676–85. [19] de Luise C, Brimacombe M, Pedersen L, Sørensen HT. Comorbidity and mortality following hip fracture: a population-based cohort study. Aging Clin Exp Res 2008;20:412–8.
5
[20] Perez JV, Warwick DJ, Case CP, Bannister GC. Death after proximal femoral fracture—an autopsy study. Injury 1995;26:237–40. [21] Harriss LR, Ajani AE, Hunt D, et al. Accuracy of national mortality codes in identifying adjudicated cardiovascular deaths. Aust N Z J Public Health 2011;35:466–76. [22] Acharya MR, Williams SC, Harper WM. Hip fracture patients: why do they die? J Bone Joint Surg [Br] 2006;88-B(Suppl I):39. [23] Wood DJ, Ions GK, Quinby JM, Gale DW, Stevens J. Factors which influence mortality after subcapital hip fracture. J Bone Joint Surg [Br] 1992;74: 199–202. [24] Yousri T, Tsai A, Thyagarajen D, Livingstone J. Medical causes of death in hip fragility fractures: mortality analysis in a primary care trauma centre. In: EFORT congress; 2011 [abstract]. [25] Cruz E, Cano JR, Benitez-Parejo N, Rivas-Ruiz F, Perea-Milla E, Guerado E. Age as a risk factor of nosocomial infection after hip fracture surgery. Hip Int 2010;20(Suppl 7):19–25. [26] National Confidential Enquiry into Patient Outcome and Death. Knowing the risk—a review of the peri-operative care of surgical patients, http://www.ncepod.org.uk/2011report2/downloads/POC_summary.pdf (date last accessed 15 January 2013). [27] Gupta BP, Kirkland LL, Huddleston JM, Huddleston PM, Burton MC, Dirk LR, et al. Risk factors for long-term mortality in very elderly after post-operative myocardial infarction following hip fracture surgery. J Am Coll Cardiol 2011;57:E1038. [28] Finnes TE, Meyer HE, Falch JA, Medhus AW, Wentzel-Larsen T, Lofthus CM. Secular reduction of excess mortality in hip fracture patients >85 years. BMC Geriatrics 2013;13:25. [29] Cornwall R, Gilbert MS, Koval KJ, Strauss E, Siu AL. Functional outcomes and mortality vary among different types of hip Fractures: a function of patient characteristics. Clin Orthop Relat Res 2004;425:64–71. [30] Damhuis RAM, Wijnhoven BPL, Plaisier PW, Kirkels WJ, Kranse R, van Lanschot JJ. Comparison of 30-day, 90-day and in-hospital postoperative mortality for eight different cancer types. Br J Surg 2012;99:1149–54. [31] Hahnel J, Burdekin H, Anand S. Re-admissions following hip fracture surgery. Ann R Coll Surg Engl 2009;91:591–5.
Please cite this article in press as: Khan SK, et al. The risk of cardiorespiratory deaths persists beyond 30 days after proximal femoral fracture surgery. Injury (2014), http://dx.doi.org/10.1016/j.injury.2014.02.024
